Re: fsck seg fault failure on vmware -i386?

[Eduardo Horvath <eeh%NetBSD.org@localhost>]

On Sun, 24 Jan 2010, yancm%sdf.lonestar.org@localhost wrote:

> > On Sun, 24 Jan 2010, David Holland wrote:
> >
> >> > Is there any other NetBSD-Fu I can use to fix this simply? 8-)
> >>
> >> Ooh boy. No, but we don't like to see fsck crashing. Are you willing
> >> to try debugging it complicatedly?
> >
> > Wouldn't this be a good time to learn about mounting the file system
> > via rump?  At the very least, your whole system won't crash, and it
> > might be easier to debug with gdb than with ddb.
> >
> > :)
> 
> Paul and David,
> 
> Sure, will take a shot at diagnosis with whatever tools and time I have.
> I've used gdb in some light debugging, never meddled in ddb that I'm aware.
> 
> Point me to docs and/or give me a rough series of commands / objectives
> and I'll give it a try.

You shouldn't need to do anything with the kernel, just gdb running on 
fsck_ffs.  You probably need to recompile fsck_ffs with symbols.  

OTOH this is probably a lost case.  While fsck_ffs should not be 
generating segmentation faults, even if it were fixed it probably wouldn't 
be able to repair your filesystem.  ffs carefully schedules writes to make 
sure imporant information has hit non-volatile storage so fsck can recover 
things if the OS crashes.  But if you're running in a virtualized 
environment and some of the data the guest thinks has been written to 
stable storage is actually cached in RAM by the host and was lost, fsck 
may not be able to fix the problem.

Eduardo